DeepSeek’s revolutionary new AI completely upset the U.S. tech stock market recently.
It was previously believed that training and developing new AI models cost upwards of $100 million.
But then China went ahead and released a model as powerful, intelligent, and fast as ChatGPT or any other American AI — but with just $6 million.
As a result, U.S. stock prices plummeted and trillions were lost in valuation overnight.
Tech stocks are only now starting to recover from their major collapse.
But, it won’t be long until the next collapse comes…
And I have reason to believe this one could have far more devastating consequences for your portfolio.
You see, while it may now be “cheaper” to train and develop new AIs…
It still demands a resource that we simply cannot supply enough of:
Energy.
Operating and developing new artificial intelligence models requires unparalleled energy demands… energy demands we cannot currently meet.
As MIT Technology Review exposed, “training a single AI model can emit as much carbon as five cars in their lifetimes.”
That’s not operating… just development.
And right now, thousands upon thousands of AI models are being trained every single day.
That’s why, by 2027, AI servers are predicted to consume as much as 134 terawatt hours annually.
In other words, in less than two years, AI will have the same annual energy consumption as countries like Argentina, the Netherlands, and Sweden.
And by 2030, AI alone could demand 25% of the grid.
This explosion in energy requirements is simply not sustainable.
As Mark Zuckerberg said recently, he’d be building out many more of these colossal data centers if they “could get the energy to do it.”
The fact is, AI adoption is only at maybe 1% of where we’ll be in the next few years.
For AI to penetrate just 10% or 20% of the market, we’ll need unprecedented amounts of energy.
Elon Musk predicts that we soon won’t have “enough electricity…” and that by 2045 the power demand in the U.S. will have tripled from current levels.
Sam Altman says “an energy breakthrough is necessary for future artificial intelligence.”
Or as the Washington Post bluntly puts it: “Amid explosive [AI] demand, America is running out of power.”
These guys know, as I do, that unless the insatiable energy demands of artificial intelligence are met, the industry will never truly go mainstream.
The release of China’s DeepSeek technology has led some to believe that AI will require less computing power to train AI models… resulting in demand for electricity falling.
This is very wrong.
Strange as it sounds, the fact that DeepSeek has proven that we need less energy to train AI means EVEN MORE energy will ultimately be consumed.
This strange phenomenon is called Jevons Paradox.
No comments:
Post a Comment